Subband coding of radiographic image sequence using a neural network approach

One important class of radiographic image sequences is angiography. If these image sequences are to be stored and transmitted in PACS they need to be compressed using a proper coding technique. Here, a subband coding technique for radiographic image sequences is presented. This coding technique can be used at transmission rates from 64 Kb/s to 5.625 Mb/s and well preserves the edges in angiography. In particular, the proposed predictor can predict upper band-to-band, thus preserving the edges. This property is important with respect to the human visual system and the diagnosis.